As the Apple Turns: Scene 4169: "Whose Bench?" "Mark's." (8/27/03)

"Whose Bench?" "Mark's." (8/27/03)

By now everyone's sick to death of all that "Apple cheated on the G5 benchmarks" / "No they didn't" hoo-haa. (Well, everyone but The Register, apparently; faithful viewer David Poves notes that they appear to be reporting on the original Apple-published benchmarks now, despite the fact that those numbers have been around since June. Let's do the Time Warp again!) Our final word on the subject was essentially that everyone should wait until the real, honest-to-Steveness production G5s hit the streets, and then we'd all see some real performance tests start to emerge. Well, guess what? The G5s are a-hittin' and the tests results are emergin'. Unfortunately, at first glance, they aren't exactly the sort of numbers to swoon over.

Let's say you flat-out ignored the first public batch of owner benchmarks, which surfaced last week over at Mac Rumors, on the grounds that they showed the G5's XBench scores barely edging out the G4's-- even after XBench had been tuned for improved G5 performance-- and therefore "couldn't be right." Okay, fine. So you decided to wait for a reasonable real-world benchmark, preferably a cross-platform one that was optimized for every processor on which it was run. Sounds like Photoshop to us! And guess what? Mac Rumors now has preliminary PS7Bench results for the G5 as well. If you're expecting to get blown away, though, it might be best if you turn away now. We hate seeing looks of disappointment.

Don't get us wrong-- the results are okay, they just aren't astounding. A handy table of all results is available at Chaosmint.com, and while the G5 turns in some pretty respectable times, it just doesn't mop the floor with the other chips as Apple would have us believe. The G5 beat out all the x86 reference systems in only 5 out of 21 tests; the 3.06 GHz Pentium 4 won 10 of the races. Granted, that doesn't mean much, so we added the times together and found that the G5 appears to have placed second to last out of seven systems, taking 180.1 seconds to complete the test battery, and finishing well behind (among other things) a 3.06 GHz Pentium 4 (124.6 seconds), a 2.8 GHz Pentium 4 (155.7) and an Athlon 2700+ (151.4).

There is a bright side, however, which is that the tested G5 was a 1.6 GHz model-- the entry-level Power Mac. Unfortunately, there's also a dark side to the bright side, which is that you can't spit without hitting a Wintel system with, say, a 2.8 GHz P4 that costs less than that entry-level Power Mac-- and that's with a 17-inch LCD display. On the lower end, at least, it's starting to look like the G5 isn't necessarily all it's cracked up to be. Ultimately, yes, benchmarks are artificial, but there's also anecdotal testimony floating around, like the words of a reseller quoted over at Accelerate Your Macintosh! who "can't say [he] see[s] a marked improvement in performance."

Our hope is that the dual 2.0 GHz G5 will show a clear performance lead at the top end, and that upcoming G5 optimizations in Panther and future releases of Photoshop only extend the G5's lead even further. Right now, though, we're definitely a little underwhelmed with the first sets of G5 benchmarks making the rounds. Perhaps the tester had the secret red toggle switch on the motherboard flipped to "SLOW"?

SceneLink (4169)

And Now For A Word From Our Sponsors

Mash-ups and original music by AtAT's former Intern and Goddess-in-Training

The above scene was taken from the 8/27/03 episode:

August 27, 2003: If you're waiting for a dual-processor G5 system, you may be waiting a little while longer-- your Mac just shipped to some school somewhere. Meanwhile, early user benchmarks of the low-end G5 are a little lackluster, and the combination of Virtual PC refusing to run on G5 systems and RealPC getting canceled because it's "vaporware" makes for a conspiracy theory that practically writes itself...

Other scenes from that episode:

4168: First Apples For Teachers (8/27/03) Still waiting for that dual 2.0 GHz Power Mac G5 you ordered about fifteen milliseconds after they appeared at the Apple Store? Get in line, pal. Oh, wait-- you placed your order, so you already are in line...
4170: And Nixon's In On It, Too (8/27/03) Conspiracy alert! Conspiracy alert! We have a Code 3 in progress! You may recall our previous conspiracy theory regarding the state of x86 emulation on the Mac: Microsoft buys the only usable such product (Virtual PC) from Connectix, and then ties up the release of the only viable competing product (RealPC) in legal red tape by issuing a cease and desist letter, thusly gaining 100% control-- temporarily-- over Mac users who need to run Windows applications...

Or view the entire episode as originally broadcast...


	As an Amazon Associate, AtAT earns from qualifying purchases


	DISCLAIMER: AtAT was not a news site any more than Inside Edition was a "real" news show. We made Dawson's Creek look like 60 Minutes. We engaged in rampant guesswork, wild speculation, and pure fabrication for the entertainment of our viewers. Sure, everything here was "inspired by actual events," but so was Amityville II: The Possession. So lighten up. Site best viewed with a sense of humor. AtAT is not responsible for lost or stolen articles. Keep hands inside car at all times. The drinking of beverages while watching AtAT is strongly discouraged; AtAT is not responsible for damage, discomfort, or staining caused by spit-takes or "nosers." Everything you see here that isn't attributed to other parties is copyright ©,1997-2024 J. Miller and may not be reproduced or rebroadcast without his explicit consent (or possibly the express written consent of Major League Baseball, but we doubt it).